[2011.07125] Better, Faster Fermionic Neural Networksopen searchopen navigation menucontact arXivsubscribe to arXiv mailings

The Fermionic Neural Network (FermiNet) is a recently-developed neural network architecture that can be used as a wavefunction Ansatz for many-electron systems, and has already demonstrated high accuracy on small systems. Here we present several improvements to the FermiNet that allow us to set new records for speed and accuracy on challenging systems. We find that increasing the size of the network is sufficient to reach chemical accuracy on atoms as large as argon. Through a combination of implementing FermiNet in JAX and simplifying several parts of the network, we are able to reduce the number of GPU hours needed to train the FermiNet on large systems by an order of magnitude. This enables us to run the FermiNet on the challenging transition of bicyclobutane to butadiene and compare against the PauliNet on the automerization of cyclobutadiene, and we achieve results near the state of the art for both.

2 mentions: @pfau@q9ac
Date: 2020/11/19 02:21

Referring Tweets

@pfau We're excited to share that our paper on making the FermiNet faster and more accurate was accepted to the NeurIPS workshop on ML for physical sciences, and we are open-sourcing the JAX code used in the paper! t.co/J75Av21LxT t.co/8IUD5x7NVI
@q9ac 多体系の波動関数近似マシーンとしてのニューラルネットワーク、最近よく見かけるなあ t.co/ehf4PozREf

Related Entries

Read more GitHub - mrlamm/textual-analogy-parsing: Code and data from the paper "Textual Analogy Parsing: What...
1 users, 6 mentions 2018/11/19 16:53
Read more GitHub - Guinsoon/KDD-CUP-2019: Context-Aware Multi-Modal Transportation Recommendation
0 users, 1 mentions 2019/07/07 03:47
Read more exBERT
0 users, 12 mentions 2019/10/15 20:18
Read more ExBERT Demo 2020 - YouTube
0 users, 1 mentions 2020/04/29 14:21
Read more https://chainer.github.io/migration-guide/#h.33jrfcvcf9zh
0 users, 1 mentions 2020/06/28 09:51